textbook-vectorization

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill uses well-known and standard Python libraries for its core functionality:
    • pypdf for text extraction from PDF files.
    • sentence-transformers for generating vector embeddings using local models.
    • numpy for mathematical operations like cosine similarity.
    • openai (optional) for cloud-based embeddings.
  • [COMMAND_EXECUTION]: The skill provides scripts that are intended for local execution by the user to process textbooks. The commands provided are standard development workflows (e.g., uv run, uv add).
  • [DATA_EXFILTRATION]: No evidence of unauthorized data transmission was found. The skill processes local PDF files and writes results to a local JSON file within the project structure. Use of the OpenAI API for embeddings is a standard and documented feature of the skill.
  • [PROMPT_INJECTION]: No prompt injection or bypass patterns were detected in the instructions or metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 01:41 AM
Security Audit — agent-trust-hub — textbook-vectorization